Apparatus for diagnosing breast cancer, the apparatus comprising a controller having a set of instructions executable to: acquire a contrast enhanced region of interest (CE-ROI) in an X-ray image of a patient's breast, the X-ray image comprising X-ray pixels that indicate intensity of X-rays that passed through the breast to generate the image; determine a texture neighborhood for each of a plurality of X-ray pixels in the CE-ROI, the texture neighborhood for a given X-ray pixel of the plurality of X-ray pixels extending to a bounding pixel radius of BPR pixels from the given pixel; generate a texture feature vector (TF) having components based on the indications of intensity provided by a plurality of X-ray pixels in the CE-ROI that are located within the texture neighborhood; and use a classifier to classify the texture feature vector TF to determine whether the CE-ROI is malignant.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
3. The method according to claim 1 and comprising processing the X-ray image to determine a contour for the CE-ROI.
This invention relates to medical imaging, specifically X-ray image processing for identifying and analyzing regions of interest in cardiac electrophysiology (CE-ROI). The problem addressed is the accurate and automated detection of cardiac regions in X-ray images, which is critical for procedures like catheter ablation where precise localization is essential. The method involves capturing an X-ray image of a patient's chest, which includes the heart and surrounding anatomical structures. The image is then processed to identify and isolate the cardiac region of interest (CE-ROI). This processing includes applying image segmentation techniques to distinguish the heart from other tissues and structures in the X-ray image. The segmentation may involve thresholding, edge detection, or machine learning-based approaches to accurately delineate the heart's boundaries. Once the CE-ROI is identified, the method further processes the X-ray image to determine the contour of the CE-ROI. This contour represents the outer boundary of the heart or a specific cardiac region, providing a precise outline for further analysis or procedural guidance. The contour determination may involve fitting mathematical models, such as splines or polygons, to the segmented region to define its shape accurately. The method ensures that the CE-ROI is accurately localized and its contour is precisely determined, enabling improved visualization and targeting in cardiac interventions. This automation reduces reliance on manual annotation and enhances procedural efficiency and accuracy.
4. The method according to claim 3 and comprising determining a contour feature vector CF based on the determined contour and concatenating the texture feature vector TF and the contour feature vector CF to provide a feature vector CF-TF.
The invention relates to image processing, specifically to a method for extracting and combining texture and contour features from an image to improve object recognition or classification. The method addresses the challenge of accurately representing both the texture and shape characteristics of objects in an image, which is crucial for tasks such as object detection, segmentation, or classification in computer vision applications. The method first involves determining a texture feature vector (TF) from the image, which captures local patterns, color variations, or other texture-based information. Additionally, a contour feature vector (CF) is derived from the image, representing the shape or boundary characteristics of objects within the image. These two feature vectors are then concatenated to form a combined feature vector (CF-TF), which integrates both texture and contour information. This combined feature vector provides a more comprehensive representation of the image, enhancing the accuracy of subsequent analysis tasks. The method may also include preprocessing steps, such as image segmentation or normalization, to improve feature extraction. The texture and contour features may be extracted using techniques such as histogram of oriented gradients (HOG), local binary patterns (LBP), or deep learning-based approaches. The concatenation of the feature vectors ensures that both texture and contour information are jointly utilized, leading to improved performance in applications like object recognition, medical imaging, or autonomous systems.
5. The method according to claim 4 and comprising using the feature vector CF-TF to determine whether the CE-ROI is malignant.
This invention relates to medical imaging analysis, specifically a method for detecting malignant regions in medical images using feature vectors. The method addresses the challenge of accurately identifying cancerous regions in medical scans, such as CT or MRI images, by leveraging machine learning techniques to analyze image features. The process involves extracting a feature vector, referred to as CF-TF, from a candidate region of interest (CE-ROI) within a medical image. This feature vector is derived from multiple image processing steps, including segmentation to isolate the CE-ROI and feature extraction to quantify relevant characteristics of the region. The extracted features may include texture, shape, intensity, or other statistical properties that distinguish malignant from benign regions. The method then uses the CF-TF feature vector to classify the CE-ROI as malignant or benign. This classification may be performed using a trained machine learning model, such as a classifier, that has been previously trained on labeled medical image data. The model evaluates the feature vector against learned patterns to determine the likelihood of malignancy. By automating the analysis of medical images and providing a quantitative assessment of malignancy risk, this method aims to improve diagnostic accuracy and efficiency in clinical settings. The approach reduces reliance on subjective visual inspection by radiologists, potentially leading to earlier and more consistent cancer detection.
6. The method according to claim 1 and comprising generating a Breast Imaging Reporting and Data System (BIRADS) Lexicon Classification System code for the CE-ROI to provide an indication as to whether the CE-ROI comprises a malignancy.
This invention relates to medical imaging analysis, specifically a method for classifying breast lesions using contrast-enhanced imaging to determine malignancy risk. The method involves analyzing a contrast-enhanced region of interest (CE-ROI) in breast tissue to generate a Breast Imaging Reporting and Data System (BIRADS) Lexicon Classification System code. This code provides an indication of whether the CE-ROI contains a malignant lesion. The process begins with acquiring contrast-enhanced imaging data of breast tissue, which may include dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) or other imaging modalities. The CE-ROI is then identified within the imaging data, representing an area of interest that may contain abnormal tissue. The method further includes extracting quantitative and qualitative features from the CE-ROI, such as enhancement patterns, kinetic parameters, and morphological characteristics. These features are analyzed using machine learning or statistical models to classify the CE-ROI according to the BIRADS lexicon, which standardizes reporting for breast imaging. The BIRADS classification system assigns a code that correlates with the likelihood of malignancy, helping radiologists and clinicians assess risk and determine appropriate follow-up actions. The method may also include comparing the generated BIRADS code with historical data or clinical guidelines to refine the classification. This approach improves diagnostic accuracy by leveraging automated analysis of contrast-enhanced imaging features, reducing subjectivity and enhancing early detection of breast cancer.
7. The method according to claim 1 wherein the wherein the classifier comprises a support vector machine and/or a neural network.
This invention relates to a machine learning-based classification system for analyzing data. The system addresses the challenge of accurately categorizing data points into predefined classes, particularly in scenarios where traditional classification methods may struggle with high-dimensional or complex datasets. The core method involves training a classifier to process input data and generate output classifications. The classifier is designed to handle diverse data types, including structured and unstructured information, by leveraging advanced machine learning techniques. The classifier may incorporate a support vector machine (SVM), which is effective for high-dimensional spaces and works by finding the optimal hyperplane that separates different classes. Alternatively, the classifier may use a neural network, which is capable of learning intricate patterns through multiple layers of interconnected nodes. The neural network can adapt to various data distributions and improve performance through iterative training. The system may also combine both SVM and neural network approaches to enhance accuracy and robustness. The method includes preprocessing the input data to prepare it for classification, such as normalization, feature extraction, or dimensionality reduction. The classifier is then trained on labeled data to learn the relationships between input features and their corresponding classes. Once trained, the classifier can predict the class of new, unseen data points. The system is particularly useful in applications requiring high precision, such as medical diagnosis, fraud detection, or quality control in manufacturing. The use of SVM and neural networks ensures adaptability to different problem domains and data characteristics.
8. The method according to claim 1 wherein the X-ray image is a spectral contrast enhanced digital mammography (SCEDM) image.
This invention relates to digital mammography, specifically enhancing image contrast using spectral techniques. The method involves capturing a spectral contrast enhanced digital mammography (SCEDM) image, which improves tissue differentiation by leveraging energy-dependent X-ray absorption properties. The process includes acquiring multiple X-ray images at different energy levels, processing these images to extract spectral information, and generating a final enhanced image that highlights subtle tissue variations. This technique addresses the challenge of low contrast in conventional mammography, particularly for dense breast tissue, by utilizing spectral differences to improve lesion detection and diagnostic accuracy. The method may also incorporate additional image processing steps, such as noise reduction and artifact correction, to further refine the output. By analyzing X-ray absorption at multiple energy levels, the technique provides a more detailed and accurate representation of breast tissue, aiding in early detection of abnormalities. The invention is particularly useful in clinical settings where high-resolution imaging is critical for diagnosing breast cancer and other conditions.
9. A non-transitory computer-readable medium comprising stored thereon executable instructions that are executable to perform the method of claim 1.
A system and method for processing data involves a non-transitory computer-readable medium storing executable instructions that, when executed, perform a data processing operation. The method includes receiving input data, analyzing the input data to identify relevant features, and generating an output based on the identified features. The analysis may involve applying one or more algorithms to extract patterns or relationships within the data. The output can be used for various purposes, such as decision-making, classification, or further processing. The system may also include additional components, such as a data storage module for storing processed data and a user interface for displaying results. The instructions are designed to ensure efficient and accurate processing of the input data, optimizing computational resources while maintaining data integrity. The method may further include error handling mechanisms to address inconsistencies or anomalies in the input data. The system is applicable in fields such as data analytics, machine learning, and automated decision-making, where efficient and reliable data processing is essential.
11. The method according to claim 10 wherein the time dependent feature comprises a rate of increase of a contrast agent taken up by tissue imaged in the CE-ROI.
This invention relates to medical imaging, specifically contrast-enhanced imaging techniques used to analyze tissue characteristics. The problem addressed is the need for improved methods to assess tissue properties by analyzing the dynamic behavior of contrast agents within regions of interest (CE-ROIs). The method involves tracking the rate of increase of a contrast agent as it is taken up by tissue during imaging. This time-dependent feature is used to derive quantitative information about the tissue, such as perfusion, vascularity, or pathological changes. The technique may be applied in various imaging modalities, including ultrasound, MRI, or CT, where contrast agents are introduced to enhance visibility of tissue structures. The method includes selecting a region of interest (ROI) within the imaged tissue, monitoring the contrast agent uptake over time, and calculating the rate of increase in contrast agent concentration within the ROI. This rate is then used to generate diagnostic or prognostic information, such as identifying abnormal tissue regions or assessing treatment response. The invention may also involve preprocessing steps like noise reduction, motion correction, or signal normalization to improve the accuracy of the rate calculation. The derived rate of increase can be displayed as a numerical value, a graphical trend, or integrated into a larger diagnostic workflow. This approach enhances the ability to detect subtle tissue changes that may not be visible in static images, improving diagnostic accuracy in conditions like cancer, ischemia, or inflammation.
12. The method according to claim 10 wherein the time dependent feature comprises a rate of decrease of a contrast agent washed out from tissue imaged in the CE-ROI.
This invention relates to medical imaging, specifically contrast-enhanced imaging techniques used to assess tissue perfusion. The problem addressed is the need for improved methods to analyze contrast agent dynamics in tissue, particularly to quantify perfusion characteristics that are clinically relevant for diagnosing or monitoring conditions like tumors, ischemia, or inflammation. The method involves tracking the time-dependent behavior of a contrast agent within a contrast-enhanced region of interest (CE-ROI) in tissue. Specifically, it measures the rate at which the contrast agent is washed out from the tissue over time. This washout rate is a key indicator of tissue perfusion, as faster washout may suggest higher blood flow or vascular permeability, while slower washout may indicate reduced perfusion or abnormal tissue characteristics. The technique may be applied in various imaging modalities, such as computed tomography (CT), magnetic resonance imaging (MRI), or ultrasound, where contrast agents are used to enhance visibility of blood flow and tissue structures. By analyzing the temporal changes in contrast agent concentration, the method provides quantitative data that can be used to assess tissue health, identify abnormalities, or monitor treatment responses. The invention improves upon existing methods by focusing on the dynamic washout phase, which offers additional diagnostic insights compared to static or peak contrast measurements.
13. The method according to claim 10 and comprising generating a Breast Imaging Reporting and Data System (BIRADS) Lexicon Classification System code for the CE-ROI to provide an indication as to whether the CE-ROI comprises a malignancy.
This invention relates to medical imaging analysis, specifically the classification of contrast-enhanced regions of interest (CE-ROIs) in breast imaging to assess malignancy risk. The method involves analyzing a CE-ROI within a breast image to determine its characteristics, such as shape, texture, and enhancement patterns, which are indicative of potential malignancy. The analysis includes comparing the CE-ROI against predefined criteria or models to assess its likelihood of being malignant. Based on this assessment, the method generates a Breast Imaging Reporting and Data System (BIRADS) Lexicon Classification System code, which provides a standardized classification of the CE-ROI. This code indicates whether the CE-ROI is likely benign, suspicious, or malignant, aiding radiologists in diagnosis. The classification system leverages machine learning or statistical models trained on annotated medical imaging data to improve accuracy. The method may also incorporate additional imaging features, such as dynamic contrast enhancement or spatial distribution, to refine the classification. The goal is to enhance early detection of breast cancer by providing automated, objective assessments of suspicious regions in medical images.
14. The method according to claim 10 wherein the wherein the classifier comprises a support vector machine and/or a neural network.
A method for classifying data using machine learning techniques addresses the challenge of accurately identifying patterns or categories within complex datasets. The method employs a classifier, specifically a support vector machine (SVM) or a neural network, to analyze input data and assign it to predefined classes. Support vector machines are supervised learning models that identify optimal hyperplanes to separate data points into distinct classes, while neural networks use interconnected layers of nodes to learn hierarchical representations of data. The classifier processes input features extracted from the data, applying learned parameters to generate classification outputs. This approach enhances decision-making by automating the categorization of data, improving efficiency and reducing human error in tasks such as image recognition, text analysis, or anomaly detection. The use of SVMs or neural networks allows the method to adapt to various data types and complexities, providing robust and scalable classification solutions. The method may also include preprocessing steps to normalize or transform input data, ensuring compatibility with the classifier's requirements. By leveraging advanced machine learning algorithms, the method offers a reliable framework for data classification across multiple applications.
15. The method according to claim 10 wherein the X-ray image is a spectral contrast enhanced digital mammography (SCEDM) image.
This invention relates to medical imaging, specifically enhancing the diagnostic accuracy of mammography by using spectral contrast-enhanced digital mammography (SCEDM). The method improves upon traditional mammography by incorporating spectral information to better differentiate between tissues, such as distinguishing between normal and abnormal breast tissue. The technique involves capturing X-ray images at multiple energy levels to generate spectral data, which is then processed to enhance contrast and highlight subtle abnormalities that may be missed in conventional single-energy mammograms. This spectral contrast enhancement helps radiologists detect early-stage tumors, microcalcifications, and other subtle features with greater precision. The method may also include image reconstruction techniques to combine the spectral data into a single high-contrast image or a series of images that provide additional diagnostic insights. By leveraging spectral information, the technique aims to reduce false positives and improve early detection of breast cancer, addressing limitations in conventional mammography where low contrast can obscure critical details. The invention is particularly useful in screening and diagnostic settings where high sensitivity and specificity are required.
16. A non-transitory computer-readable medium comprising stored thereon executable instructions that are executable to perform the method of claim 10.
A system and method for managing data processing tasks in a distributed computing environment. The technology addresses inefficiencies in task scheduling, resource allocation, and workload balancing across multiple computing nodes, leading to suboptimal performance and resource utilization. The invention provides a solution that dynamically adjusts task distribution based on real-time system conditions, such as node availability, processing capacity, and network latency, to optimize overall system throughput and reduce idle time. The method involves monitoring the status of computing nodes in a distributed network, analyzing their current workload and performance metrics, and dynamically assigning tasks to nodes based on their availability and efficiency. It includes mechanisms for load balancing, where tasks are redistributed to prevent overloading any single node while ensuring underutilized nodes are effectively utilized. The system also incorporates predictive algorithms to anticipate future workload demands and preemptively adjust resource allocation to maintain optimal performance. The non-transitory computer-readable medium stores executable instructions that implement this method, ensuring that the system can adapt to changing conditions in real-time. The instructions are designed to interface with existing distributed computing frameworks, making the solution compatible with various hardware and software configurations. This approach enhances scalability, reduces processing delays, and improves energy efficiency by minimizing unnecessary resource consumption. The invention is particularly useful in large-scale data processing environments, such as cloud computing, big data analytics, and high-performance computing clusters.
18. A non-transitory computer-readable medium comprising stored thereon executable instructions that are executable to perform the method of claim 17.
A system and method for processing data involves analyzing input data to identify patterns or anomalies. The method includes receiving input data from one or more sources, such as sensors, databases, or user inputs. The system then processes the input data to extract relevant features, such as statistical measures, temporal trends, or spatial relationships. These features are analyzed using machine learning models or statistical algorithms to detect patterns, anomalies, or other significant characteristics. The results of this analysis are then outputted, either as alerts, reports, or visualizations, to inform decision-making or further processing. The system may also include a feedback mechanism to refine the analysis based on user input or additional data. The method can be applied in various domains, such as predictive maintenance, fraud detection, or quality control, where identifying patterns or anomalies in data is critical. The system may be implemented on a computing device, including a processor and memory, where the instructions for performing the method are stored and executed. The system may also include interfaces for receiving input data and outputting results, ensuring seamless integration with existing data sources and user interfaces.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
February 1, 2021
December 27, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.